The objective of this study was to evaluate the discriminative power of different nonmotor signs for early diagnosis of Parkinson's disease (PD). Thirty patients with PD with <or=3 years of disease duration were compared with 30 healthy controls. Six deficit domains (DD) were defined: hyposmia, sleep abnormalities, dysautonomia, visual deficits, executive dysfunction, and depression. Plotting of Receiver operating characteristic (ROC) curves and exact conditional logistic modeling, followed by manual stepwise descending procedure were used to identify a model for nonmotor signs that detects early PD. Patients with PD and controls did not differ in terms of age, gender, and educational level. Several DD discriminated patients with PD from healthy controls. Visual deficits showed the largest area under the ROC curve (0.83), followed by hyposmia (0.81) and dysautonomia (0.80). When combining the DD visual deficits and dysautonomia, the best residual model was obtained; it maximized both sensitivity and specificity for PD at a level of 0.77. At an early disease stage, several nonmotor domains were already able to discriminate patients with PD from healthy controls. Visual deficits had the best discriminatory power. Being brief and inexpensive, visual tests should be further investigated in larger cohorts as potential screening tool for early PD.
(c) 2010 Movement Disorder Society.